Re: Graham released Arc

[Ed Watkeys <edw@poseur.com>]

> * Really? It's not just an eval hack?

What is the difference between "really" and "just an eval hack"?

> Slicing doesn't appear to work. i.e. you can't say ("andrew" 1 3) and  
> expect to get back "ndr", or even "nd" if the second argument isn't a  
> count...

I tried the same thing. There doesn't seem to be a reason not to do
that, aside from 1) available time and 2) it's of marginal value
compared to the rationale for the unary function. Specifically, the
big win with collections-as-functions seems to be in contexts like
this:

(= cipher "nopqrstuvwxyzabcdefghijklm")

(def letter-index (ch)
     (- (coerce ch 'int) 97))

(def rot-13 (s)
     (map cipher:letter-index s))

(rot-13 "foo") ==> "sbb"
(rot-13 "sbb") ==> "foo"
(rot-13:rot-13 "foo") ==> "foo"

Starting to wrap my head around the potential here. Anyway, the point
is, collections-as-mere-unary-functions have a lot of potential.

Regarding non-MzScheme hosted Arcs, it seems like a lot of pointless
work at this stage. A lot of Lisps have lived hosted in other Lisps
for extended periods. Arc is still subject to a lot of change. I don't
see the value of a self-hosting Arc or even an Arc that's hosted
elsewhere until it gels some more.

Ed